The Digital Past Course Blog – Page 16

Ghost Stories from the 1930’s and Voyant Tools

My historical question was “In what ways have ghost stories changed, if at all, and based on these accounts are people more or less careful around ghosts? Why?” and while I was curious about the answer, I had figured that it would probably be similar to the ghost stories we tell today, although I did expect people to care about it less, which ended up being untrue from what I read.

When I first started looking up words, I wasn’t sure if I was going to find any, but luckily there were a good number of actual ghost stories. I used words like “ghost” “spirit” and “supernatural” with more results than I was expecting, although sometimes I’d get more religious texts with “spirit”, or people would use phrases like “pale as a ghost” which was not what I was looking for.

It took me a bit to really understand Voyant and which tool would work best for me. Since my topic is pretty niche, I found that the word map wasn’t helpful, but that the graph and the contrast tools helped me compare my corpus with the class’ to find similarities and differences ¹.

The amount of times the word “ghost” showed up in my corpus

The amount of times the word “ghost” showed up in the class’ corpus

What I did in this picture was I entered the word “ghost” into both my corpus and the class’ corpus to see how often the word shows up throughout texts. Most of the spikes in the class’ corpus was from the stories I submitted, but there was one that was very high that I hadn’t submitted. It ended up being just a sentence from an interview but because the interview was only one sentence, it made the graph think that because “ghost” was a common word in the interview, simply because there were few other words, that means it must be relevant.

In the end I feel like my historical question doesn’t fit what I found so I’d change it to something more like “How common were ghost stories? How important are they to the communities their told in?” because I found that they were somewhat common throughout many towns and they inspired bonding in certain stories ². I also found that I enjoyed looking through the different ghost stories and how they ended up being similar to the ones we tell today, despite how different the times are. One of the most interesting things I found was how the debate over whether ghosts are real or not was just as relevant then as it is now, with both skeptics and believers arguing. The biggest difference I think is how today, people mostly talk about evidence being caught on camera, and whether or not it’s faked, but in the 1930’s, catching a ghost on camera would be near impossible so the beliefs were based on personal experience. The conversations have shifted from personal to objective, unaltered proof, which while can be more believable, it discounts a lot of experiences that cannot be seen, such as the emotional energy tied to a place, and credible psychics, however rare they might be.

Cultural Identity In Folklore

My original historical question when approaching this project was, “How is cultural identity shared through folklore?” Initially, when exploring the folklore unit of the FWP Life Histories, I expected to find a large array of stories passed down from different cultures. This expectation was bolstered by the numerous Life Histories categorized by ancestry, and a number of interviewees of different ancestries, aside from “American” and “American Negro.”

My first step in analyzing the texts was to limit the corpus to documents only from the folklore unit of the FWP Life Histories. This entailed modifying the corpus to keep only documents that included the term “folklore” (the interviewers organized them by listing “folklore” under the subject.) This limited the corpus to exactly 100 documents.

Using Voyant was overwhelming at first, but after just experimenting with each tool for a while, I learned how to specifically use them. I found that for the purposes of my historical question, simply uploading the corpus and glancing at the trending terms was relatively useless.

The original results of the Cirrus tool, showing the trending words from the entire corpus, reveal mostly generic words, most of them regarding the form the interviewers filled out. Further inspection was necessary in order to yield results relevant to my question regarding cultural identity.

Applying stopwords globally³ helped to narrow down the results by excluding terms and phrases that were being counted due to the formatting of the documents (i.e “http,” “Library of Congress,” “ww.loc.gov,” ” Ò ,” etc.).

It became evident soon though that my idea of cultural folklore was not what was being expressed through the life histories interviews. Rather than old wives’ tales and bedtime stories reflecting ancestral heritage, I found everyday stories about America that just so happened to be told by subjects of several different ancestries. This revelation was reached by using the trends tool to search for several different ancestries (German, Dutch, Irish, Yiddish, Jewish, Russian). This made it easier to read a select number of documents closely and find that they were pretty typical accounts of day to day life at the time, lives that were simply American lives.

Searching explicitly for the ancestries I had seen through close reading revealed how often they were recorded. Despite my perception that there had been many interviews of subjects of Irish descent, there was a greater amount of interviewees of Jewish descent.

These revelations of how simplistically cultural identity was reflected in day to day life reflects Botkin’s motivations in assembling” ethnically diverse life histories… to foster the tolerance necessary for a democratic, pluralistic community.” ⁴ Overall, this analysis reveals that while lives are experienced differently based on ancestry and cultural uniquities, there was certainly a common experience for many in America at the time despite diversity, that hopefully was a unifying factor.

Blog Post 6

How did the hope and hard-work of everyday Americans pull us out of the Depression? I asked this question because we have been talking about how the Federal Arts Project was what really helped us out of the Great Depression but I was curious to see how much the everyday working Americans really contributed. My initial thought was that the working class was the main reason we survived the Depression. Without the working class to work all the jobs provided by the Federal Arts Project, the projects would not have succeeded. So, I believe that the sole reason America made it through the Great Depression, was because of its everyday working-class Americans. When selecting my sources, I looked for ones that talked about work and providing for their family. I searched using a few different filters. I began with words such as: work, family, living, and hard. Then choose the type of source I wanted: manuscript. I used these filters and found my ten manuscripts that I believed would help me to answer my historical question.

I am still deciding whether I like Voyant tools or not.⁵ When I first started using it I kept getting frustrated with it. I didn’t know how to do certain things and it was really confusing. Once I sat down, read a bit about it, and played around with it, I began to understand it more. ⁶ I was able to compare three documents at once which was helpful. After messing around with the program, I began to analyze the texts. When comparing the texts, I could narrow it down by searching for certain words such as: unemployed. This was helpful to search through other people’s documents that they found. After using Voyant Tools and analyzing several different texts, the conclusions I came too were equivalent to my initial ideas. I found that, in fact, it was the hard work of the Americans that fueled the Federal Arts Projects, that then pulled America out of the Great Depression. As far as text analysis goes, it was harder than I had expected. It took me a while to find sources that I could later analyze and it took a while for me to get accustomed to Voyant Tools.

Mining the Eleanor Roosevelt My Day Columns, regan008.shinyapps.io/mining_my_day/.

Miriamposner. “Miriamposner/Voyant-Workshop.” GitHub, github.com/miriamposner/voyant-workshop/blob/master/investigating-texts-with-voyant.md.

The American Life Histories Collection and Florida

My historical question was, “How did the Great Depression impact farmers in Florida?”. I was for sure expecting to find texts in the “American Life Histories” collection related to Florida and farming. I assumed that I will be able to read scripts from interviews of farmer’s life and working conditions during the Great Depression.

Of course, I first made sure that my research was within the collection. Then I filtered my research by state, and input “Florida”. There were only a few texts related to Florida which made it easy to read through and pick the texts I wanted for my corpus.

I had the chance to use a software called “Voyant Tools”. This is a great solution to analyze huge amounts of data and practice distant reading. It is also great to correlate with close reading thanks to the tools offered on the platform.

I believe that Cirrus tool was the most useful one. It allows me to see the most used words in the corpus. You can also add stop words to delete from it, analyze any words you do not wish to see, and leave space for others which might be more important.

WordTree is a great tool too. Indeed, it highlights different words and also gives a notion of hierarchy between the words.

I first decided to upload the entire corpus. Cirrus tool (see below) gave a huge diversity of words. I had to use the tool Terms. I went back to close reading and picked few words to put in the search bar of Cirrus Show Terms. The most relevant words were: Farm (x591), and Farming (x64); Life*(x749) and Work* (x2773); Child* (x641) and Family* (x616). We can see that the stories are more about the life at the farm than farming. Stories talk about the life of people, their work, and family. The distant reading confirms the trend I noticed from my close reading.

The first analyzation is not enough to bring any answers to my historical question. Because I focus on Florida, I decided to reduce the corpus to all texts from Florida. It is a corpus of 42 documents (as opposed to 323 documents before).

I used Cirrus tool (see below) but added the following stop words: resource, congress, www.loc.gov, http, library, got. The word Florida shows up and this is reassuring. The analysis is working. The following important words appeared: work, good, home, Florida, mother, house, children, people, school, home. They are all related to my historical question.

Using the tool Terms, I compared the Florida corpus within the entire corpus using a proportion scale for each interesting word: Work (22%), Child (26%), Family (25%), Farm (38%), and Farming (44%).

The word “farm” is in a proportion of 38% within the Florida corpus out of the entire corpus. This only means that texts related to farming and farm within the entire corpus are mostly from the Florida texts.

There is definitely a trend for family, child, and work.

Then I ran WordTree (see below). The following words are organized around the word good: farm, strawberry, time, farmer.

“Good” seems to lead to a positive atmosphere related to the farm, the time spent, the life of the farmer, and surprisingly strawberries. It is confirmed by close reading, “as it is, we have good land and good health, plenty of good wholesome food, a roof over our heads, and good children” ⁷

I was expecting to get a deep understanding of the working conditions of farmers in Florida and a description of their work. Nevertheless, Benjamin Botkin said that he wanted to move “the streets, the stockyards, and the hiring halls into literature.” In other words, we are looking at the everyday life of people. The text analysis reflects it. We are actually not looking at their working conditions but at their own life, feelings, and what later on they believed was relevant to share from this period with the interviewer.

I paired my findings with close readings in order to understand why certain words would come up more than others. I also picked words from close reading to inject them in Voyant to see where it would lead my research.

After using Voyant Tools, I understood that I should modify my historical question. I did not find any useful information around farming within the collection. I should instead focus on “how the life of Florida people was impacted by the Great Depression?”.

Text analysis is important for digital historians in order to better understand the past. It is essential to pair distant and close readings. Indeed, using one or the other alone could lead to non-sense. I found interesting to cross data and come up with an interpretation to answer my historical question.

Blog Post #6

My historical question was: How did the theatre impact the morale of people of color in the late 1930s? Originally, I felt like I could answer my own question with research and in-class readings about how it made theatre more accessible to all, more diverse, and lifted the spirits/served as escapism of the Great Depression

⁸

. However, after realizing that I wouldn’t find a personal narrative to say that, it made this assignment more difficult. My question was refined, changed, and even entirely overhauled days before the assignment was due but, what I have now, I really like.

My searching process was very specific. I started with narrowing down the date. Next, I started to do some keyword searching. I searched “black theatre” “black theater” “people of color performance” “colored theatre” “theatre and race” “theater and race” “black actress” “black actor” “federal theater project black” and “black plays”. I came across some really good documents about black actors being interviewed about how their life changed while on Broadway or vaudeville and others about the parties people threw afterwards. To me, it seemed clear that the general disposition was lifted. I picked sources that talked about the personal experiences of black actors. Finding these sources was harder than I thought but when I did find them, it was exactly what I was looking for.

I liked Voyant tools but I don’t think I will need to use it again. It’s very specific for this type of at-a-distance reading. As a theater major, I probably could use it to analyze scripts. But, for what it’s used for, it was really useful. My favorite tools were Cirrus and TermsBerry because they showed exactly what I wanted, frequency. I narrowed down the stopwords that I needed to get rid of that our tutorial so graciously taught us about

⁹

I compared my specific documents to the corpus as a whole to better understand how they differed from everyday rhetoric to the niche rhetoric of performance. I did a good job of exploring the different tools that Voyant had to offer. But ultimately, the research I have struggled to answer my question. So I did a little more closer reading to my texts and tried to understand if those readings confirmed my hypothesis. All of my readings centered around how much people of color were happy when talking about being in theatre which confirmed my concerns with my historical question. In the end, I feel like my question was answered. I was shown that people of color had a better outlook on their lives after the Federal Theater Project facilitated them being involved in the theatre. I was definitely searching for specific documents that answered my historical question how I wanted to see it and not how it objectively appeared but, I don’t yet have the corpus to truly delve into the truth of my historical question which, can only be left up to the future.

Blog Post 6: Unions & Text Analysis

My original historical question was, “what factors helped shift the public perception of labor unions in the 1930s, and how?” And with “public perception,” I was referring to the working class’ trust in unionization and the efficacy of unions. Initially, I assumed that labor unions were stigmatized negatively in the 1930s because of their lack of success in years prior, and that affected the willingness of the working class to unionize.¹⁰

I selected my sources from the American Life Histories collection by refining the search with the subject of “labor unions.” The items I selected for part one were particularly interesting to me because of the passion and candor the interviewees spoke with when referring to unionization; there was actually a document that was a manuscript of a Worker’s Alliance Meeting in New York, and the entire thing was just the prayer that someone led before starting the meeting.

Voyant turned out to be really beneficial for text-analysis and helping me create a more constructive question. I utilized the class corpus by creating a new corpus with only 62 documents that involved the words “union” or “unionization.” Distant reading was useful in this aspect because it allowed me to see the frequency in which unions were talked about throughout the collection. Now that I had a corpus that was composed of manuscripts about labor unions, I wanted to see how people were talking about them.

In the “contexts” section of the site, I was able to search “union good”~15 and “union bad”~15, and this allowed me to find sentences with the words “union” and “good” within 15 words of each other. I got a total of ten results and only 1/10 of those documents used the word “bad” to describe unions in a negative manner. The other nine documents used the word “good” to refer to unions in a positive light. After pairing my findings with close reading, I realized that the documents do not clearly and explicitly present the factors that affect the working class’ perspective on labor unions. So, my original historical question went from, “what factors helped shift the public perception of labor unions in the 1930s, and how?” to “what was the public perception of labor unions in the 1930s?”

Searching *“union bad”~15* in the corpus of 62 documents.

To conclude, the manuscripts in my close reading of the 10 documents proved to be useful in emphasizing peoples candid thoughts and passion about labor unions. But most importantly it shed light on how there were discrepancies in attitudes about unionization, that were determined primarily by your status as a working-class individual. For example, the manuscripts showed that while some people had the freedom to unionize, others were fearful of retaliation in the workplace. Also, despite reading Miriam Posner’s blogpost beforehand, I still went into this assignment thinking that Voyant would provide a “straightforward window into a text.”¹¹ But by the end of the assignment, I realized, as Posner argues, that text analysis and Voyant should instead be used “as a way to talk about how we interact with texts and what kind of meaning evades capture by computational tools.”¹²

Look to the Margins

In the FWP’s records how does the involvement and voice of African-American Women vary through singular vs. familial accounts?

Do African-American Women fall into the abyss of words and accounts as the corpus of FWP focuses on other voices?

I selected my sources by submitting myself to the mercy of the Library of Congress digital algorithms. After I used the African-American filter to ensure I was in the right place I picked the first 5 women’s accounts and the first 5 family/couples accounts. I did this because someone wanting to see the basic African-American women’s account would see these documents. So I purposefully did not use severe discretion with my options.

The Voyant tools were incredibly useful in thier scope in their use of distant reading. Uploading the corpus was easy and efficient. Of all the tools the “Distinct Word” and “Vocabulary Density” feature stood out the most. Voyant alligns the documents in terms of most which ones have the most dense vocabulary vs. the least dense vocabulary. The women’s and Familial accounts both have pretty evenly dense vocabulary, and the same ratio is in the document length.

These comparisons show that with this document sample can truly only tease certain questions. However, those questions it teases are lost within the confines of an overflow of text in the broad corpus.

My question is about the voices of African-American Women, and there is no way to that within the confines of this overwhelming amount of documents that the voice of a Minority can be heard. If I am to properly seek an answer to my question my 10 documents need to be the topic of study.

Everyone during the great depression was concerned about work. That concern is reflected in the common word in all data clumps of the word “work.” However, One particular feature that stands out among the comparison are the dialect words.

In the African-American women’s accounts dialect words are a huge component of the culture. One story in particular, [I Got a Record], records not only what she says but how she says it. Dialect history is as much as necessary part of history as any.

In the main corpus many of the dialect words are not present, and thus erases an element of the story. Their culture is lost in a see of words. Their stories are just so unique and different that it does not “fit” with the rest of storylines. For history to be told properly it needs to be organized according to catagory. Rather than an overarching narative filling the space a need for categorized and harmonious naratives are needed. Every topic must have a consideration of race and gender because it changes the perameters.

Historical questions are only as good as the tactics we use. If we wish to look at the broad picture of the entire corpus, then we see the main concerns of the entirety of the people. But with this decision lose certain aspects of it within the quantity of words. Distent reading is good for answering questions of generality, but close reading is necessary to understand the specific categorical naratives.

Depression-Era Religion

My historical question was: “How did the role of religion within American society shape and be shaped by the social climate of the 1930s, and were peoples’ religious beliefs strengthened or weakened by the hardships of the Depression?”

Going into this assignment, I already knew that the late 1920s saw a rise in conservatism, especially with the advent of the radio paving the way for pre-televangelism. This took societal hold even further in the 1930s, and I always assumed that was owed to people seeking a higher power to turn to during the Depression.

When selecting my sources, I knew that I wanted to focus on religion: prayer, church, priests. But when one of my sources wound up being a man speaking of his discussion with a Jew on the rise of fascism in Europe, that got me interested in Judaic aspects as well, instead of keeping the question fully focused on Christian reactions to the Depression.

This got even more interesting with Voyant: ‘church’ and ‘god’ are, perhaps unsurprisingly, the most common terms out of my searches. But ‘jew’ far surpasses ‘priest’, ‘jesus’, and ‘christian’. I can think of an explanation for ‘christian’ being so low – almost all the speakers from the sources that I read ensured to emphasize their denomination of Christianity, while that isn’t as much of an issue in Judaism. But even after I tried to filter for terms like ‘catholic’, ‘protestant’, or ‘evangelical’, the numbers were still much lower.

Out of curiosity, I came back to the source that inspired me to include Judaism, and I searched for the term ‘fascism’. It appears eight times – six of which come from that source. The Jewish man within it speaks of how he fears that only other Jews cared about what was happening to Europe in 1939. I don’t think this is unfounded. American goyim either didn’t seem to care or were convinced that it wouldn’t spread. ‘Fascism’ only overlaps with one document containing the Christianity-related terms that I searched for, and the word is scarce in mention across the rest of the corpus. True, America was dealing with its own problems, but nothing happens in a vacuum.

In this vein, I might choose to revise my question into, “How did religious beliefs and differences affect the concerns of America during the Depression?” Many of the Christian sources speak of attempts to convert more sheep to the flock, or prescribed prayer in the hopes of a divine remedy for domestic strife. Christianity seems to have been confined to two spheres: home and church. Judaism, while more centralized in urban/immigrant-heavy places like New York City, takes to the streets and gathering places like cafes. As someone who comes from both Catholic and Jewish backgrounds, I definitely find it fitting that the Christian sources are actively attempting to solve what they perceive as “problems”, while the Jewish sources pose more questions.

Musicians’ Homes in the 1930s

My historical question was: How were the lives of musicians in the 1930s, primarily minorities, impacted by the time period/events? After reading through many interviews that seemed like a great question to answer, however, after using the voyant tool and looking at these interviews through a distant reading scale, I was able to notice things I didn’t before. I was initially able to see what words were most commonly used through the use of the “word cloud, with more frequently occurring words shown in a larger size.”¹³ The first word that stuck out to me was “home.” It’s a simple word that holds variable meanings depending on who’s using it.

My question shifted from how musicians’ lives were impacted, to what did these musicians consider to be their home? To stay on the theme of musicians I was limited to using only my documents, as the others did not quite apply. This proved to work out better anyways as it allowed me to better compare how words were used in fewer documents. One of the features I relied on was term searching and plotting them on the trend graph. My initial goal was to search between terms I thought would relate to “home” and see if the trends aligned.

While the meaning of “home” clearly means different things to the different people in each interview, there are still some similarities. The greatest similarity between all individuals that were interviewed, is that they used the words “home” or “house” to a similar frequency they did as words used to describe family members. With a few exceptions, physical places weren’t commonly used, so it can be assumed, many of them didn’t associate home with a physical location. This makes sense when considering the migratory nature of the time period, especially with so many musicians being inspired by the events of the dust bowl and great depression.¹⁴¹⁵ Although through distant reading, it can be difficult to tell for certain whether or not this is entirely accurate, it provides a great start to look at these interviews more closely and gain a better understanding.

Text analysis

For this project, I decided to use my own corpus, which is the 10 files that relate directly to my historical question. My historical question is, “How did the aftermath of the Great Depression affect the people who lived during it”. Through this question, I was able to find several files that matched up well with this topic and gave me a great insight into my question. However, I chose to compare it to the class corpus as well to try and be and visualize the similarities and differences between both of the corpus.

The American Yawp gave me great insight into the great depression and how it affected everyone’s lives. They talked about how it affected the market directly by saying, ” The market temporarily stabilized but fears spread over the weekend and the following week frightened investors dumped their portfolios to avoid further losses.”. ¹⁶ “The Great Depression” American yawp. http://www.americanyawp.com/text/23-the-great-depression/¹⁷. This quote helped me narrow down what I wanted to do for my distant learning even before I had used Voyant tools. The podcast on Hoover, who was president at the beginning of the great depression, talked about what happened to people after the market crashed. “Nearly half of its banks failed, nearly a quarter of its workforce out of a job.”. ¹⁸ “The Memory Palace”, Podcast, http://thememorypalace.us/2017/10/hoover/ . This led me to decide what I wanted to look at in my corpus. What I immediately noticed within my corpus was the word business. It was easily the biggest word on the Cirrus image. I was able to find that the word business was mentioned over 100 times in all of my files; this led me to do my distant learning based on this word because all of the other words were smaller and did not hold as much importance. When looking through where the word business fell in my files I found a pattern. Every time the word was mentioned it was either that the person owned a business, worked within a government entered business, or just were apart of a standard business. Several people had lost or closed their business as soon as the stock market crash while others did a few years later. For example, in one of my files titled “Honest and Fairness to the Bitter End”, a man had to close his business in 1936, seven years after the stock market crash. He had mentioned that people have tried to help him keep it open, but nothing was successful. I was surprised that he was even able to keep it open for a whole seven years. Secondly, I saw that many people had mentioned that after the stock market crash the government promised that businesses would resume but most of them never did.

My last step was comparing my 10 files based on my historical question to the large class corpus. First off, there was a larger amount of results in the class corpus because there were 300 more files. I was surprised to see that in the larger corpus I could not even find the word business in the cirrus. I thought I would for sure see the word business but I could not find any information on it. The biggest word in the larger corpus was “resources” and it was mentioned almost 2000 times; compared to the largest word in my corpus which was only mentioned about 100 times. Secondly, the trend graph for both of the corpus was drastically different. The class corpus trend graph was much more congested and harder to visualize. Whereas my corpus trend graph was easier to understand and pick out certain themes. In conclusion, I was able to compare both of the corpus and how we can learn to distant learn from a little information to a larger amount of information.

Citations: “The Great Depression” American yawp. http://www.americanyawp.com/text/23-the-great-depression/

“The Memory Palace”, Podcast, http://thememorypalace.us/2017/10/hoover/.